Detail-revealing Deep Video Super-resolution
This addresses the problem of enhancing video resolution for applications like media and surveillance, representing an incremental improvement over existing CNN-based approaches.
The paper tackles video super-resolution by introducing a sub-pixel motion compensation layer in a CNN framework to improve frame alignment, resulting in high-quality outputs that outperform state-of-the-art methods without parameter tuning.
Previous CNN-based video super-resolution approaches need to align multiple frames to the reference. In this paper, we show that proper frame alignment and motion compensation is crucial for achieving high quality results. We accordingly propose a `sub-pixel motion compensation' (SPMC) layer in a CNN framework. Analysis and experiments show the suitability of this layer in video SR. The final end-to-end, scalable CNN framework effectively incorporates the SPMC layer and fuses multiple frames to reveal image details. Our implementation can generate visually and quantitatively high-quality results, superior to current state-of-the-arts, without the need of parameter tuning.